Neurogenetic algorithm applied to Route Planning for Autonomous Mobile Robots

Abstract

We developed a bioinspired algorithm to assist in the navigation of an autonomous mobile robot in dynamic environments. The robotic controller uses both an Artificial Neural Network (ANN) and a Genetic Algorithm (GA), aided by a low computational cost vision system. The controller uses the vision system and the ANN to detect and recognize obstacles found in the robot's path. If the object is in the controller's knowledge bank a previously registered deviation solution is applied. Otherwise, the GA must optimize a new route alternative. We modeled and simulated the controller using robot's simulator V-REP, and the Computer Vision System using Scilab software. The contribution of this paper is the development of a hybrid neuro-genetic algorithm to control the navigation of autonomous mobile robots in dynamic environments.